Impact on excitations
Temperature can first influence electromagnetic excitations (e.g. slotting harmonics) responsible for magnetic noise in electric machines. It is already known that cogging torque varies with temperature due to reduced remanent flux of permanent magnets in synchronous machines.
Temperature can affect the electrical circuit resistances, and depending on the control it may lead to current variations. Stator current increase will increase magnetic forces induced by stator field, resulting in potential higher noise.
It is therefore difficult to give a general trend on force harmonics with temperature.
Impact on excited structure
Temperature can affect the contact properties of the mechanical assembly (especially the shrink fit between stator and frame), and the resulting natural frequencies. It can also affect the resin properties and the stiffness of the copper/resin mixture in stator slots, resulting in lower natural frequencies and higher damping when temperature increases.
Application to e-NVH
As temperature affects both electromagnetic excitations and excited structure, it is particularly important to measure it during experimental tests. Comparing different data tests and e-motor designs in terms of e-NVH should include thermal aspects.
In case of a small thermal constant compared to the run-up duration, NVH test repeatability can be low. In this particular case, one may have to wait for the thermal steady state before running e-NVH meaurements.
To evaluate the effect of temperature on structural modes, an Operational Modal Analysis can be performed. Alternatively, impact hammer test on stator samples in a temperature-control chamber can be carried. This characterization can be run by EOMYS testing services.
Application to Manatee
In Manatee e-NVH software, a fixed temperature is defined for the equivalent circuit and remanent flux of PM-based machines, but it can be varied to study the impact on magnetic noise.
Modal damping is not modified with the user-defined simulation temperature, a constant value is used by default but it can be varied manually using experimental data.